It was never about the stack.

A Snowflake warehouse stood up by a vendor but stalled — dev/staging/prod schemas built but not in active use, Fivetran connections defined but not turned on. Multi-year survey data spanning four to six successive instrument versions sitting in Qualtrics. Salesforce capturing partnership phases, consultant caseloads, and assessment-cycle scheduling. Legacy operational data in Azure. A theory-of-change framework leadership already used to talk about impact.

A semantic dataset across multi-year surveys. An executive analytics surface for program operations. An impact-reporting workflow that mapped data back to the theory-of-change — instead of being authored alongside it. A targeted-funding view — surfacing where burden and disparity diverge across states, sized against public-health datasets.

Activated the stalled warehouse with DevOps — turned on the Fivetran connections, validated freshness, brought the dev/staging/prod schemas into analytical use. Designed the semantic dataset architecture for multi-year survey analysis so the data team could build the report-ready layer underneath. Stood up a ThoughtSpot trial as a demonstration vehicle for executive dashboards on Salesforce program operations and national dataset DWH analytics — proof of value ahead of any procurement decision. Brought CDC Wonder data into the warehouse — moving multi-year trend and subgroup analyses from piece-meal Google Sheets queries into DWH-based workflow. Extended the same data into a targeted-funding view — a burden-vs-disparity quadrant by state, framing where need and underinvestment co-occur for grant-strategy use. Prototyped AI-assisted impact reporting using Snowflake Cortex, Streamlit, Gemini, and Python — mapping outputs to the theory-of-change domains so impact narrative could be drafted from data instead of authored separately.

Three reporting platforms running in parallel — Looker, Tableau Cloud, Tableau Public. Tableau Cloud licenses on the books for two years, mostly unused. Student-level academic dashboards on public infrastructure with no governance. Daily attendance reporting that duplicated what the SIS already showed. Google Sheets workflows running quietly under most decisions. HubSpot, SchoolMint, PowerSchool, NWEA MAP, Illuminate, Google Forms, MailChimp, and a call center workflow — all real, all uneven.

A consolidated, governed reporting environment. A semantic data foundation that didn't depend on a missing BigQuery ETL layer. Dashboards refocused from daily attendance to chronic absenteeism, persistence, and school-health trends. An enrollment funnel principals and ops teams could act on through the recruitment cycle.

Authored a data maturity assessment and phased roadmap (Stage 0 Siloed → Stage 1 Developing → Stage 2 Predictive). Consolidated three platforms into Tableau Cloud, operationalizing the licenses already on the books. Migrated student-level dashboards off Tableau Public for governance. Built the semantic data foundation on Google Sheets → Tableau, working around the missing BigQuery ETL. Drove enrollment forecasting at 99% accuracy via ARIMA — correcting an inherited attrition-rate error in the Looker model. Modernized recruitment from ad-hoc tabling into a measurable acquisition funnel integrating HubSpot, SchoolMint, PowerSchool, lead scoring, zipcode-based digital marketing, MailChimp, and the call center workflow with tiered ops/enrollment-team follow-ups.

Validated psychometric instruments (PHQ-9, GAD-7, C-SSRS Adult and Child, NIDA Quick Screen, CAGE-AID, plus SIS, PAQ, Q-LES-Q, FIBSER) historically captured at 3-month visits only. A smartphone app being piloted to collect these as patient-reported every 2 to 4 weeks instead — supporting measurement-based care between visits. 837i claims feeds across multiple payers (Beacon Health, Fidelis Care NY, Health First, Partners Health Plan, United Healthcare) sitting separately. HL7/CCDA hospital encounter feeds from regional EDs, uncoordinated. Program operations data across Article 31, OnTrack NY, and Health Home. HR and financials in their own systems. Demographic and zipcode-level data the org had access to but wasn't analyzing.

A HIPAA-compliant patient-360 foundation. Real-time care coordination off ED and hospital encounter feeds. Instrument-level dashboards back in front of clinicians. A measurement-based care workflow that closed the loop after the visit. SDOH and chronic-condition risk scoring. NSDUH state-level prevalence integrated into population-health analytics.

Architected a HIPAA-compliant Snowflake EDW POC with two contract engineers, integrating clinical assessments, 837i claims, HL7/CCDA hospital encounter feeds via Mirth Connect (real-time care coordination alerts), referrals, ED utilization, Article 31 / OnTrack NY / Health Home program operations, HR and financials, and demographic + zipcode data into a unified patient-360 analytical foundation. Built Tableau dashboards across the validated instruments, alongside operational and financial dashboards on caseload by clinician and managing office, claims by clinician, 837i payer submissions, ED utilization across regional hospitals, 30-day re-admission alerts, and SDOH + chronic-condition risk scoring. Designed the Mitram measurement-based care workflow with Otsuka — a pilot with 30+ patients across two waves drove +40% engagement after reminder automation and follow-ups. Integrated NSDUH prevalence data into the population-health view.

A legacy SQL Server environment with 300+ ad-hoc requests for reports and insights in the queue. Three analytics FTEs plus roughly 10% of 30+ business-side FTEs spending time on manual data wrangling. A 55% annual reporting completion rate. A standing 8–12-request weekly inventory. Booking data flowing through six channels (Internet, Inbound Call, Local Office, Enrollment Advisor, Service Center, Marketing) across SAT/ACT/MCAT/GRE/GMAT/LSAT product lines. Score-gain data sitting unused for cross-tier analysis. Stakeholder voices across Product, Ops, Marketing that hadn't been collected into a single proposal.

A BI platform of record. Repeatable booking analytics across six product lines and six channels. Efficacy research the marketing and product teams could actually use. Cross-functional decision support across A/B testing, customer profiling, cancel/refund analysis, scheduling, and location-level enrollment.

Authored an Analytics & Reporting 2.0 strategic proposal — diagnosed the legacy environment, captured stakeholder voices, ran a six-vendor BI evaluation (Tableau, Spotfire, Sisense, Domo, Periscope, Looker), and presented tiered investment scenarios. Selected Sisense as the platform of record. Implemented Sisense ElastiCube ETL across booking analytics for six product lines — daily/monthly pacing, budget-vs-actual, YoY trends, channel attribution, product mix. Conducted efficacy research on test-prep products — score-gain analyses across SAT, GRE, MCAT by instructor, location/PSO, and product tier. Delivered cross-functional decision support: A/B testing, high-value customer profiling, cancel/refund driver analysis, promotional-campaign performance, scheduling optimization, location-level enrollment data for leasing decisions.

Existing platform log data. Course-level student outcome data on the university side — traditional and hybrid sections of the same course running in parallel. Stakeholders thinking about a next-generation test-prep product without a measurement frame.

A defensible set of log-data measures. A platform-design frame the product team could build against. A defensible analytical frame for comparing student outcomes between traditional and hybrid sections of the same course.

Designed log-data measures and contributed to the NextGen test-prep platform design. Built conceptual dashboard designs as simulated charts — enough fidelity for product and stakeholder decisions, without waiting for the warehouse that wasn't built yet. On the university side: student outcomes analyses between traditional and hybrid sections of the same course.